Search in sources :

Example 1 with TokensStartState

use of org.antlr.v4.runtime.atn.TokensStartState in project antlr4 by antlr.

the class ATNDeserializer method deserialize.

@SuppressWarnings("deprecation")
public ATN deserialize(char[] data) {
    data = data.clone();
    // was implemented.
    for (int i = 1; i < data.length; i++) {
        data[i] = (char) (data[i] - 2);
    }
    int p = 0;
    int version = toInt(data[p++]);
    if (version != SERIALIZED_VERSION) {
        String reason = String.format(Locale.getDefault(), "Could not deserialize ATN with version %d (expected %d).", version, SERIALIZED_VERSION);
        throw new UnsupportedOperationException(new InvalidClassException(ATN.class.getName(), reason));
    }
    UUID uuid = toUUID(data, p);
    p += 8;
    if (!SUPPORTED_UUIDS.contains(uuid)) {
        String reason = String.format(Locale.getDefault(), "Could not deserialize ATN with UUID %s (expected %s or a legacy UUID).", uuid, SERIALIZED_UUID);
        throw new UnsupportedOperationException(new InvalidClassException(ATN.class.getName(), reason));
    }
    boolean supportsPrecedencePredicates = isFeatureSupported(ADDED_PRECEDENCE_TRANSITIONS, uuid);
    boolean supportsLexerActions = isFeatureSupported(ADDED_LEXER_ACTIONS, uuid);
    ATNType grammarType = ATNType.values()[toInt(data[p++])];
    int maxTokenType = toInt(data[p++]);
    ATN atn = new ATN(grammarType, maxTokenType);
    //
    // STATES
    //
    List<Pair<LoopEndState, Integer>> loopBackStateNumbers = new ArrayList<Pair<LoopEndState, Integer>>();
    List<Pair<BlockStartState, Integer>> endStateNumbers = new ArrayList<Pair<BlockStartState, Integer>>();
    int nstates = toInt(data[p++]);
    for (int i = 0; i < nstates; i++) {
        int stype = toInt(data[p++]);
        // ignore bad type of states
        if (stype == ATNState.INVALID_TYPE) {
            atn.addState(null);
            continue;
        }
        int ruleIndex = toInt(data[p++]);
        if (ruleIndex == Character.MAX_VALUE) {
            ruleIndex = -1;
        }
        ATNState s = stateFactory(stype, ruleIndex);
        if (stype == ATNState.LOOP_END) {
            // special case
            int loopBackStateNumber = toInt(data[p++]);
            loopBackStateNumbers.add(new Pair<LoopEndState, Integer>((LoopEndState) s, loopBackStateNumber));
        } else if (s instanceof BlockStartState) {
            int endStateNumber = toInt(data[p++]);
            endStateNumbers.add(new Pair<BlockStartState, Integer>((BlockStartState) s, endStateNumber));
        }
        atn.addState(s);
    }
    // delay the assignment of loop back and end states until we know all the state instances have been initialized
    for (Pair<LoopEndState, Integer> pair : loopBackStateNumbers) {
        pair.a.loopBackState = atn.states.get(pair.b);
    }
    for (Pair<BlockStartState, Integer> pair : endStateNumbers) {
        pair.a.endState = (BlockEndState) atn.states.get(pair.b);
    }
    int numNonGreedyStates = toInt(data[p++]);
    for (int i = 0; i < numNonGreedyStates; i++) {
        int stateNumber = toInt(data[p++]);
        ((DecisionState) atn.states.get(stateNumber)).nonGreedy = true;
    }
    if (supportsPrecedencePredicates) {
        int numPrecedenceStates = toInt(data[p++]);
        for (int i = 0; i < numPrecedenceStates; i++) {
            int stateNumber = toInt(data[p++]);
            ((RuleStartState) atn.states.get(stateNumber)).isLeftRecursiveRule = true;
        }
    }
    //
    // RULES
    //
    int nrules = toInt(data[p++]);
    if (atn.grammarType == ATNType.LEXER) {
        atn.ruleToTokenType = new int[nrules];
    }
    atn.ruleToStartState = new RuleStartState[nrules];
    for (int i = 0; i < nrules; i++) {
        int s = toInt(data[p++]);
        RuleStartState startState = (RuleStartState) atn.states.get(s);
        atn.ruleToStartState[i] = startState;
        if (atn.grammarType == ATNType.LEXER) {
            int tokenType = toInt(data[p++]);
            if (tokenType == 0xFFFF) {
                tokenType = Token.EOF;
            }
            atn.ruleToTokenType[i] = tokenType;
            if (!isFeatureSupported(ADDED_LEXER_ACTIONS, uuid)) {
                // this piece of unused metadata was serialized prior to the
                // addition of LexerAction
                int actionIndexIgnored = toInt(data[p++]);
            }
        }
    }
    atn.ruleToStopState = new RuleStopState[nrules];
    for (ATNState state : atn.states) {
        if (!(state instanceof RuleStopState)) {
            continue;
        }
        RuleStopState stopState = (RuleStopState) state;
        atn.ruleToStopState[state.ruleIndex] = stopState;
        atn.ruleToStartState[state.ruleIndex].stopState = stopState;
    }
    //
    // MODES
    //
    int nmodes = toInt(data[p++]);
    for (int i = 0; i < nmodes; i++) {
        int s = toInt(data[p++]);
        atn.modeToStartState.add((TokensStartState) atn.states.get(s));
    }
    //
    // SETS
    //
    List<IntervalSet> sets = new ArrayList<IntervalSet>();
    // First, read all sets with 16-bit Unicode code points <= U+FFFF.
    p = deserializeSets(data, p, sets, getUnicodeDeserializer(UnicodeDeserializingMode.UNICODE_BMP));
    // deserialize sets with 32-bit arguments <= U+10FFFF.
    if (isFeatureSupported(ADDED_UNICODE_SMP, uuid)) {
        p = deserializeSets(data, p, sets, getUnicodeDeserializer(UnicodeDeserializingMode.UNICODE_SMP));
    }
    //
    // EDGES
    //
    int nedges = toInt(data[p++]);
    for (int i = 0; i < nedges; i++) {
        int src = toInt(data[p]);
        int trg = toInt(data[p + 1]);
        int ttype = toInt(data[p + 2]);
        int arg1 = toInt(data[p + 3]);
        int arg2 = toInt(data[p + 4]);
        int arg3 = toInt(data[p + 5]);
        Transition trans = edgeFactory(atn, ttype, src, trg, arg1, arg2, arg3, sets);
        //			System.out.println("EDGE "+trans.getClass().getSimpleName()+" "+
        //							   src+"->"+trg+
        //					   " "+Transition.serializationNames[ttype]+
        //					   " "+arg1+","+arg2+","+arg3);
        ATNState srcState = atn.states.get(src);
        srcState.addTransition(trans);
        p += 6;
    }
    // edges for rule stop states can be derived, so they aren't serialized
    for (ATNState state : atn.states) {
        for (int i = 0; i < state.getNumberOfTransitions(); i++) {
            Transition t = state.transition(i);
            if (!(t instanceof RuleTransition)) {
                continue;
            }
            RuleTransition ruleTransition = (RuleTransition) t;
            int outermostPrecedenceReturn = -1;
            if (atn.ruleToStartState[ruleTransition.target.ruleIndex].isLeftRecursiveRule) {
                if (ruleTransition.precedence == 0) {
                    outermostPrecedenceReturn = ruleTransition.target.ruleIndex;
                }
            }
            EpsilonTransition returnTransition = new EpsilonTransition(ruleTransition.followState, outermostPrecedenceReturn);
            atn.ruleToStopState[ruleTransition.target.ruleIndex].addTransition(returnTransition);
        }
    }
    for (ATNState state : atn.states) {
        if (state instanceof BlockStartState) {
            // we need to know the end state to set its start state
            if (((BlockStartState) state).endState == null) {
                throw new IllegalStateException();
            }
            // block end states can only be associated to a single block start state
            if (((BlockStartState) state).endState.startState != null) {
                throw new IllegalStateException();
            }
            ((BlockStartState) state).endState.startState = (BlockStartState) state;
        }
        if (state instanceof PlusLoopbackState) {
            PlusLoopbackState loopbackState = (PlusLoopbackState) state;
            for (int i = 0; i < loopbackState.getNumberOfTransitions(); i++) {
                ATNState target = loopbackState.transition(i).target;
                if (target instanceof PlusBlockStartState) {
                    ((PlusBlockStartState) target).loopBackState = loopbackState;
                }
            }
        } else if (state instanceof StarLoopbackState) {
            StarLoopbackState loopbackState = (StarLoopbackState) state;
            for (int i = 0; i < loopbackState.getNumberOfTransitions(); i++) {
                ATNState target = loopbackState.transition(i).target;
                if (target instanceof StarLoopEntryState) {
                    ((StarLoopEntryState) target).loopBackState = loopbackState;
                }
            }
        }
    }
    //
    // DECISIONS
    //
    int ndecisions = toInt(data[p++]);
    for (int i = 1; i <= ndecisions; i++) {
        int s = toInt(data[p++]);
        DecisionState decState = (DecisionState) atn.states.get(s);
        atn.decisionToState.add(decState);
        decState.decision = i - 1;
    }
    //
    if (atn.grammarType == ATNType.LEXER) {
        if (supportsLexerActions) {
            atn.lexerActions = new LexerAction[toInt(data[p++])];
            for (int i = 0; i < atn.lexerActions.length; i++) {
                LexerActionType actionType = LexerActionType.values()[toInt(data[p++])];
                int data1 = toInt(data[p++]);
                if (data1 == 0xFFFF) {
                    data1 = -1;
                }
                int data2 = toInt(data[p++]);
                if (data2 == 0xFFFF) {
                    data2 = -1;
                }
                LexerAction lexerAction = lexerActionFactory(actionType, data1, data2);
                atn.lexerActions[i] = lexerAction;
            }
        } else {
            // for compatibility with older serialized ATNs, convert the old
            // serialized action index for action transitions to the new
            // form, which is the index of a LexerCustomAction
            List<LexerAction> legacyLexerActions = new ArrayList<LexerAction>();
            for (ATNState state : atn.states) {
                for (int i = 0; i < state.getNumberOfTransitions(); i++) {
                    Transition transition = state.transition(i);
                    if (!(transition instanceof ActionTransition)) {
                        continue;
                    }
                    int ruleIndex = ((ActionTransition) transition).ruleIndex;
                    int actionIndex = ((ActionTransition) transition).actionIndex;
                    LexerCustomAction lexerAction = new LexerCustomAction(ruleIndex, actionIndex);
                    state.setTransition(i, new ActionTransition(transition.target, ruleIndex, legacyLexerActions.size(), false));
                    legacyLexerActions.add(lexerAction);
                }
            }
            atn.lexerActions = legacyLexerActions.toArray(new LexerAction[legacyLexerActions.size()]);
        }
    }
    markPrecedenceDecisions(atn);
    if (deserializationOptions.isVerifyATN()) {
        verifyATN(atn);
    }
    if (deserializationOptions.isGenerateRuleBypassTransitions() && atn.grammarType == ATNType.PARSER) {
        atn.ruleToTokenType = new int[atn.ruleToStartState.length];
        for (int i = 0; i < atn.ruleToStartState.length; i++) {
            atn.ruleToTokenType[i] = atn.maxTokenType + i + 1;
        }
        for (int i = 0; i < atn.ruleToStartState.length; i++) {
            BasicBlockStartState bypassStart = new BasicBlockStartState();
            bypassStart.ruleIndex = i;
            atn.addState(bypassStart);
            BlockEndState bypassStop = new BlockEndState();
            bypassStop.ruleIndex = i;
            atn.addState(bypassStop);
            bypassStart.endState = bypassStop;
            atn.defineDecisionState(bypassStart);
            bypassStop.startState = bypassStart;
            ATNState endState;
            Transition excludeTransition = null;
            if (atn.ruleToStartState[i].isLeftRecursiveRule) {
                // wrap from the beginning of the rule to the StarLoopEntryState
                endState = null;
                for (ATNState state : atn.states) {
                    if (state.ruleIndex != i) {
                        continue;
                    }
                    if (!(state instanceof StarLoopEntryState)) {
                        continue;
                    }
                    ATNState maybeLoopEndState = state.transition(state.getNumberOfTransitions() - 1).target;
                    if (!(maybeLoopEndState instanceof LoopEndState)) {
                        continue;
                    }
                    if (maybeLoopEndState.epsilonOnlyTransitions && maybeLoopEndState.transition(0).target instanceof RuleStopState) {
                        endState = state;
                        break;
                    }
                }
                if (endState == null) {
                    throw new UnsupportedOperationException("Couldn't identify final state of the precedence rule prefix section.");
                }
                excludeTransition = ((StarLoopEntryState) endState).loopBackState.transition(0);
            } else {
                endState = atn.ruleToStopState[i];
            }
            // all non-excluded transitions that currently target end state need to target blockEnd instead
            for (ATNState state : atn.states) {
                for (Transition transition : state.transitions) {
                    if (transition == excludeTransition) {
                        continue;
                    }
                    if (transition.target == endState) {
                        transition.target = bypassStop;
                    }
                }
            }
            // all transitions leaving the rule start state need to leave blockStart instead
            while (atn.ruleToStartState[i].getNumberOfTransitions() > 0) {
                Transition transition = atn.ruleToStartState[i].removeTransition(atn.ruleToStartState[i].getNumberOfTransitions() - 1);
                bypassStart.addTransition(transition);
            }
            // link the new states
            atn.ruleToStartState[i].addTransition(new EpsilonTransition(bypassStart));
            bypassStop.addTransition(new EpsilonTransition(endState));
            ATNState matchState = new BasicState();
            atn.addState(matchState);
            matchState.addTransition(new AtomTransition(bypassStop, atn.ruleToTokenType[i]));
            bypassStart.addTransition(new EpsilonTransition(matchState));
        }
        if (deserializationOptions.isVerifyATN()) {
            // reverify after modification
            verifyATN(atn);
        }
    }
    return atn;
}
Also used : ArrayList(java.util.ArrayList) UUID(java.util.UUID) Pair(org.antlr.v4.runtime.misc.Pair) InvalidClassException(java.io.InvalidClassException) IntervalSet(org.antlr.v4.runtime.misc.IntervalSet)

Example 2 with TokensStartState

use of org.antlr.v4.runtime.atn.TokensStartState in project antlr4 by antlr.

the class LexerATNFactory method createATN.

@Override
public ATN createATN() {
    // BUILD ALL START STATES (ONE PER MODE)
    Set<String> modes = ((LexerGrammar) g).modes.keySet();
    for (String modeName : modes) {
        // create s0, start state; implied Tokens rule node
        TokensStartState startState = newState(TokensStartState.class, null);
        atn.modeNameToStartState.put(modeName, startState);
        atn.modeToStartState.add(startState);
        atn.defineDecisionState(startState);
    }
    // INIT ACTION, RULE->TOKEN_TYPE MAP
    atn.ruleToTokenType = new int[g.rules.size()];
    for (Rule r : g.rules.values()) {
        atn.ruleToTokenType[r.index] = g.getTokenType(r.name);
    }
    // CREATE ATN FOR EACH RULE
    _createATN(g.rules.values());
    atn.lexerActions = new LexerAction[indexToActionMap.size()];
    for (Map.Entry<Integer, LexerAction> entry : indexToActionMap.entrySet()) {
        atn.lexerActions[entry.getKey()] = entry.getValue();
    }
    // LINK MODE START STATE TO EACH TOKEN RULE
    for (String modeName : modes) {
        List<Rule> rules = ((LexerGrammar) g).modes.get(modeName);
        TokensStartState startState = atn.modeNameToStartState.get(modeName);
        for (Rule r : rules) {
            if (!r.isFragment()) {
                RuleStartState s = atn.ruleToStartState[r.index];
                epsilon(startState, s);
            }
        }
    }
    ATNOptimizer.optimize(g, atn);
    return atn;
}
Also used : TokensStartState(org.antlr.v4.runtime.atn.TokensStartState) LexerAction(org.antlr.v4.runtime.atn.LexerAction) RuleStartState(org.antlr.v4.runtime.atn.RuleStartState) Rule(org.antlr.v4.tool.Rule) HashMap(java.util.HashMap) Map(java.util.Map)

Aggregations

InvalidClassException (java.io.InvalidClassException)1 ArrayList (java.util.ArrayList)1 HashMap (java.util.HashMap)1 Map (java.util.Map)1 UUID (java.util.UUID)1 LexerAction (org.antlr.v4.runtime.atn.LexerAction)1 RuleStartState (org.antlr.v4.runtime.atn.RuleStartState)1 TokensStartState (org.antlr.v4.runtime.atn.TokensStartState)1 IntervalSet (org.antlr.v4.runtime.misc.IntervalSet)1 Pair (org.antlr.v4.runtime.misc.Pair)1 Rule (org.antlr.v4.tool.Rule)1